f77 v1.3.1

[monk at math.udel.edu]

I am having a strange problem with sun f77 v1.3.1.  Can anyone help?

Here is the code:

      double precision d1,d2,d3
      data d1,d2,d3 /0.0d0,0.1d0,0.0d0/
      write(*,*) d1,d2,d3
      stop
      end

and here are some results of running it on my machine
(sparc1, OS4.1, f77v1.3.1, local disks):

33 % f77 -O -o test test.f
34 % test
   -2.3191199484774+265    1.0000000000000D-01  0.
35 % f77 -O -o test test.f ; test
test.f:
 MAIN:
    6.3660116631392-314    1.0000000000000D-01  0.
36 % !!
f77 -O -o test test.f ; test
test.f:
 MAIN:
    2.5463949492077-313    1.0000000000000D-01  0.
37 % !!
f77 -O -o test test.f ; test
test.f:
 MAIN:
  0.    1.0000000000000D-01  0.


As you can see I can get results ranging from correct to almost correct to
terrible depending on when I run it.  Sometimes it will give the correct
result for a number of compilation/run cycles.

Is this a known bug? Is there a fix?  Any help would be greatly
appreciated.

Peter Monk